Apply Blunt and Karpicke’s format/recall separation (online 2014-02-17). A representation test would keep reviewed wording, sources, source visibility, practice and AI access fixed, with comparable time allowances, varying one feature: division into topics or prerequisite-link navigation. Do not equalize navigation when testing navigation. A whole-workflow test would compare the complete authoring, separate topic/link review, disagreement handling, learning and upkeep process with a strong reusable-document or wiki workflow supported by AI, counting all work. Matching reviewed content cannot identify review’s benefit; a whole-workflow difference cannot isolate representation. The 2014 short-text, one-week question outcomes do not establish scientific-oversight gains. These are proposed comparisons of 1Cademy’s untested workflow (undated page; captured 2026-11-09).